Single crystal scintillation is a very simple but also very sensitive method of detecting high energy radiation such as x-rays, gamma-rays and high energy particles with energies exceeding a few kilo-electron volt (KeV). In the past century, a large number of crystals have been proposed for potential scintillating applications. For medical imaging such as positron emission tomographs (PET), crystals with the highest light yield, narrowest energy resolution and fastest decay time are required. Moreover, PET also requires a crystal with good physical integrity and chemical inertness. Thallium-activated sodium iodide, NaI(Tl) has by far the highest light yield of 38,000 photons per million-electron volt (MeV). Unfortunately, NaI(Tl) is hygroscopic which is not a desirable property in making the small detector pixels used in a PET system.
The first crystal used in a PET application was bismuth germanate (BGO), which has the general formula Bi4Ge3O12. BGO is non-hygroscopic and has good physical properties. The problem of BGO is the relatively low light yield of 6,000 photons per MeV, or approximately 15% of the yield of NaI(Tl). BGO also has a long decay time of 300 nano-seconds (ns), which is too slow for the coincident detection employed in a PET system.
A much improved crystal for PET application was developed in the early 1980s. This is cerium doped gadolinium orthosilicate (GSO), which has the general formula Ce:Gd2SiO5. GSO exhibits the much shorter decay time of 60 ns and a light yield of approximately 10,000 photons per MeV, or 25% that of NaI(Tl).
In the early 1990s, cerium doped lutetium orthosilicate (LSO), having the general formula of Ce:Lu2SiO5, was discovered. LSO had, by far, may have the best overall properties. The decay time of LSO is only 47 ns and the light yield is 29,000 photons per MeV, or approximately 76% that of NaI(Tl). Even though LSO has excellent properties, it is not without problems. One of the most serious problems of LSO is the large variation of light yield from crystal boule to crystal boule and even from top to bottom within the same crystal boule. In the past, this variation has been attributed to impurities within the crystal, which generate color centers and thus quenche the radiative emission.
More recently, Chai et al. in U.S. Pat. No. 6,624,420 described the newest entry in scintillator crystals for PET, that is lutetium yttrium orthosilicate (LYSO) having the general composition of Ce2x(Lu1-yYy)2(1-x)SiO5 where x=0.00001 to 0.05 and y=0.0001 to 0.9999. LYSO also has very high light yield up to 96% that of NaI(Tl) and a similar fast decay time of 48 ns. At the same time, however, it may also suffer the same problem as LSO, that is, a large light yield variation from crystal boule to crystal boule, as well as from the top to the bottom of the same crystal boule, although the variation is much smaller than that of LSO. The smaller variation in light yield properties of LYSO was attributed to the substitution of yttrium, which has much higher purity than the lutetium source.
A large variation in light yield presents a serious problem for the construction of PET detector blocks. Ideally, it is preferred that all pixels within a detector block have the same light yield and energy resolution. Moreover, the spread of the light, or energy resolution is another important parameter in the design of the PET detector blocks. Ideally, it is preferable to have the energy resolution as narrow as possible. At the present time, neither BGO nor GSO suffers from the problem of non-uniformity in light yield and energy resolution. However, for both LSO and LYSO, these are persistent problems.
In the past, in order to provide PET detector blocks using either LSO or LYSO and having uniform performance, it would be necessary to carefully measure the light yield performance and energy resolution of each individual detector pixel. By manually selecting the pixels with nearly the same performance, it would be possible to build detector blocks having uniform performance. This is a very costly manufacturing process, however. To make the product competitive, it became necessary to develop a process that could eliminate this manual selection process and still identify all the detector pixels having the same performance.
Applicant believes, without wishing to be bound thereto, that the poor light yield performance of LSO is due to deep traps that take the energy away and dissipate it non-radiatively. However, very little is known about the exact nature of these non-radiative recombination centers. Since the starting material, Lu2O3, is only 99.95% in purity, it is believed that poor light yield performance was due to the impurities in the starting material from which the crystal is made. However, repeated chemical analyses of trace elements in both the Lu2O3 powder starting material and the whole crystal have failed to identify the exact impurity that quenches the light yield. Applicant has also intentionally added trace amounts of impurities in the starting powder mixture and then grown a LSO or LYSO crystal, but this approach failed to show any of the intended results of light quenching.
Even though both LSO and LYSO have the problem of serious light yield variation, Applicant has not found the same effect in cerium doped yttrium orthosilicate single crystal (YSO) having the general formula Ce:Y2SiO5. Moreover, while LSO has a strong afterglow which can last many hours after being radiated with UV light, Applicant has never observed a similar afterglow of YSO under the same UV radiation conditions. Since the starting material, Y2O3, for YSO is 99.999% in purity, the evidence tended to support the speculation that the light yield variation is due to the impurity of the Lu2O3 raw material.
In the development of the LYSO crystal, Applicant has also noticed the enhancement of light yield shown by LYSO over LSO, as well as much weaker afterglow for LYSO crystals with high yttrium content. At that time, this observation also supported the assumption that a high yttrium content LYSO crystal starts with less Lu2O3 and, therefore, has less impurities and thus better light yield.
Despite the seeming consistency of this pattern as indicating that impurities in the Lu2O3 raw material were the main cause of light yield reduction, the impurity or impurities which interfere with light yield has never been identified, even though Applicant has used a variety of Lu2O3 sources from many different vendors having different types and levels of impurities. In fact, the performance of both LSO and LYSO seems to be independent of the source of Lu2O3. Even when the same batch of chemicals is used, it is still possible to have a large variation in light yield from crystal boule to crystal boule. Given those circumstances, Applicant began to suspect that the cause of light yield variation is not really due to the impurity effect but due to other causes that are more fundamental and most likely related to the basic structure of the crystal.
YSO, LSO and LYSO have the same crystal structure, which is monoclinic with a space group of C2/c. The structure has two distinct rare earth cation sites. One is a distorted 7-fold coordinate site and the other one is a smaller distorted 6-fold coordinate site. These two sites are quite different from each other, with distinct energy levels for emission. When the crystal is doped with cerium, the dopant substitutes into both sites. The exact distribution ratio between the two sites is not known. However, since the emission spectra of Ce:LSO (FIG. 1) and Ce:YSO (FIG. 2) are not quite the same, it may be assumed that the Ce distribution between the two sites is different for Ce:LSO and Ce:YSO. For Ce:LYSO, because the crystal has a very high content of Lutetium (at least approximately 95%) and low Yttrium content (approximately less than 5%) the absorption and emission spectra are substantially the same as that of pure LSO.
Single crystals of YSO, LSO and LYSO are all produced by the Czochralski melt pulling technique as known in the art, since all three compositions melt congruently. However, the melting points of all three crystals are quite high, 1980° C., 2150° C. and 2100° C., respectively. To hold the molten charges at such high temperatures, it is necessary to use an iridium metal crucible as the container, which has a melting point of 2450° C. Even though iridium is quite inert and stable, it does oxidize in air at such high temperatures. To prevent metal loss of the iridium crucible, it is necessary to grow the crystals in either a vacuum or in furnaces purged with argon or nitrogen gas, so that the ambient oxygen in the growth chamber is kept below approximately 1%.
Even though the crystals grown in this method are colorless and transparent, Applicant theorizes without wishing to be bound thereto that there is some evidence that oxide crystals produced at such high temperatures under a low oxygen condition tension can generate oxygen vacancy point defect centers. These points defect centers could act as recombination centers that take away the radiative energy from an otherwise normal electron-hole recombination process. Even though such oxygen point defect centers are well known, they have not been implicated directly as the possible cause for low light yield in scintillating crystals.
During Applicant's extensive experience growing crystals of LSO, YSO and LYSO over the years, there have been a number of times that accidentally an air leakage developed in a growth chamber. This was highly undesirable, since the iridium crucible was badly burned to a purplish black color. The crystal surface, in those instances, was covered with small single crystal flakes of iridium metal released from the crucible. Generally, the process must be stopped right away to minimize further damage to the crucible and to the furnaces. In most of these cases a partially completed crystal is the result.
In these cases, however, even though LSO, YSO and LYSO all have the same crystallographic structure, the crystals produced under such highly oxygenated conditions are quite different from each other. For YSO, the crystal turns into a light yellow color, indicating oxidation of Ce from the 3+ to the 4+ state. Nevertheless, there is improvement of the light yield in these oxygenated crystals as compared with the regular YSO crystals.
For both LSO and LYSO crystals, the change from colorless to yellow is rarely seen. Most of these crystals are still transparent and colorless, but among these crystals, applicant noticed a definite improvement in light yield. When grown using the prior process of reduced oxygen, the typical light yield for an LSO crystal is about 4× that of BGO, but occasionally a crystal having 5× BGO light yield may be obtained, although rarely. Surprisingly, however, Applicant discovered that these accidentally air-leaked LSO crystals generally showed 5× or even better light yield over BGO, which is exceptionally good. For LYSO crystals, the results were even better. It was possible to obtain as much as 6× or more the light yield of BGO. Moreover, the improvement appeared to be independent of the source of Lu2O3. In other words, Applicant theorizes without wishing to be bound thereto that the improvement is unrelated to the degree of original impurities found in the crystal, or to the reduction thereof.
Because of the result of these air-leaked crystals, applicant has tried to increase the oxygen content in the growth chamber to improve the light yield. Unfortunately, applicant is not able to increase the oxygen too much without worrying about the damage of the Iridium crucible. The net result of these crystals with increasing oxygen content did not show any significant improvement in light yield within the experimental variation range. So applicant concluded that increased oxygen during growth is not an effective way to oxygenate the crystal. This result is totally consistent with the phase diagram illustrated in FIG. 11 as will be discussed in greater detail below.
Still the result of these air-leaked crystals shows for the first time a direct linkage between the light yield and the chemical nature of the crystal, i.e., crystal with exposure of high ambient oxygen during cooling has better light yield. On the other hand, applicant did not see any direct linkage of a specific impurity (within the impurity range of the normal 99.95% pure Lu2O3) to the light yield, nevertheless, and applicant can not rule out that high impurity within the crystal can also damage the light yield performance.